rinku 2.0.2 → 2.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 2982203836f1449fe5344a32f7ab1782f3cb4cc7
-  data.tar.gz: 8461bfe12528f128cba4e81e9fb8f64e9903e6e5
+  metadata.gz: efae79dff80a684e71b33c625f2335fe4867302d
+  data.tar.gz: 4b8daa1d4afe0d14a46003eb2a9e71248a283b9b
 SHA512:
-  metadata.gz: 3587f9e83406d7829cdbec671980536803f3b9e0ed87e6daf8e21389dad8d2cc521d49527b602e3ccaa7e40f3c41ef365ba4bbb42a0b69d6ed7c412bbdd914f3
-  data.tar.gz: 68aa755836af8269edc44bcd4ae8134abe6a3bd7836140218f4bd618069f22ad16ff78c6814610f2525ca5f64cfac8657e2021c0ee238e1ba5d319e6efeb728c
+  metadata.gz: 0b01768b18a0eda0cb7c2863af5fd150dbe121d4d349350fab023a29955b275dc7f2026ad3bdd69f28e688f0bc5581884c8c9d668ff3782360784779bbeeb916
+  data.tar.gz: aed5796d6929e1ee7fe111d6ff618cb216071c9fde4fa655a6a14ad1c14ccb0d5e7695e92ea9a1eb4cb917161e58fbe8b26832b69e1bf53211dd06a761d74308

data/ext/rinku/autolink.c

@@ -52,7 +52,7 @@ autolink_issafe(const uint8_t *link, size_t link_len)
 static bool
 autolink_delim(const uint8_t *data, struct autolink_pos *link)
 {
-	uint8_t cclose, copen = 0;
+	int32_t cclose, copen = 0;
 	size_t i;
 
 	for (i = link->start; i < link->end; ++i)
@@ -88,15 +88,8 @@ autolink_delim(const uint8_t *data, struct autolink_pos *link)
 	if (link->end == link->start)
 		return false;
 
-	cclose = data[link->end - 1];
-
-	switch (cclose) {
-	case '"':	copen = '"'; break;
-	case '\'':	copen = '\''; break;
-	case ')':	copen = '('; break;
-	case ']':	copen = '['; break;
-	case '}':	copen = '{'; break;
-	}
+	cclose = utf8proc_rewind(data, link->end);
+	copen = utf8proc_open_paren_character(cclose);
 
 	if (copen != 0) {
 		/* Try to close the final punctuation sign in this link; if
@@ -124,16 +117,21 @@ autolink_delim(const uint8_t *data, struct autolink_pos *link)
 		size_t i = link->start;
 
 		while (i < link->end) {
-			if (data[i] == copen)
+			int32_t c = utf8proc_next(data, &i);
+			if (c == copen)
 				opening++;
-			else if (data[i] == cclose)
+			else if (c == cclose)
 				closing++;
-
-			i++;
 		}
 
-		if (closing > opening)
-			link->end--;
+		if (copen == cclose) {
+			if (opening > 0)
+				utf8proc_back(data, &link->end);
+		}
+		else {
+			if (closing > opening)
+				utf8proc_back(data, &link->end);
+		}
 	}
 
 	return true;

data/ext/rinku/utf8.c

@@ -75,6 +75,30 @@ int32_t utf8proc_next(const uint8_t *str, size_t *pos)
 	return read_cp(str + p, length);
 }
 
+int32_t utf8proc_back(const uint8_t *str, size_t *pos)
+{
+	const size_t p = *pos;
+	int8_t length = 0;
+
+	if (!p)
+		return 0x0;
+
+	if ((str[p - 1] & 0x80) == 0x0) {
+		(*pos) -= 1;
+		return str[p - 1];
+	}
+
+	if (p > 1 && utf8proc_utf8class[str[p - 2]] == 2)
+		length = 2;
+	else if (p > 2 && utf8proc_utf8class[str[p - 3]] == 3)
+		length = 3;
+	else if (p > 3 && utf8proc_utf8class[str[p - 4]] == 4)
+		length = 4;
+
+	(*pos) -= length;
+	return read_cp(&str[*pos], length);
+}
+
 size_t utf8proc_find_space(const uint8_t *str, size_t pos, size_t size)
 {
 	while (pos < size) {
@@ -93,7 +117,7 @@ int32_t utf8proc_rewind(const uint8_t *data, size_t pos)
 	if (!pos)
 		return 0x0;
 
-	if ((data[pos - 1] & 0xC0) == 0x0)
+	if ((data[pos - 1] & 0x80) == 0x0)
 		return data[pos - 1];
 
 	if (pos > 1 && utf8proc_utf8class[data[pos - 2]] == 2)
@@ -106,6 +130,24 @@ int32_t utf8proc_rewind(const uint8_t *data, size_t pos)
 	return read_cp(&data[pos - length], length);
 }
 
+int32_t utf8proc_open_paren_character(int32_t cclose)
+{
+	switch (cclose) {
+	case '"': return '"';
+	case '\'':  return '\'';
+	case ')': return '(';
+	case ']': return '[';
+	case '}': return '{';
+	case 65289: return 65288; /* （） */
+	case 12305: return 12304; /* 【】 */
+	case 12303: return 12302; /* 『』 */
+	case 12301: return 12300; /* 「」 */
+	case 12299: return 12298; /* 《》 */
+	case 12297: return 12296; /* 〈〉 */
+	}
+	return 0;
+}
+
 bool utf8proc_is_space(int32_t uc)
 {
 	return (uc == 9 || uc == 10 || uc == 12 || uc == 13 ||

data/ext/rinku/utf8.h

@@ -26,8 +26,11 @@ bool rinku_isalpha(char c);
 bool rinku_isalnum(char c);
 
 int32_t utf8proc_rewind(const uint8_t *data, size_t pos);
+int32_t utf8proc_next(const uint8_t *str, size_t *pos);
+int32_t utf8proc_back(const uint8_t *data, size_t *pos);
 size_t utf8proc_find_space(const uint8_t *str, size_t pos, size_t size);
 
+int32_t utf8proc_open_paren_character(int32_t cclose);
 bool utf8proc_is_space(int32_t uc);
 bool utf8proc_is_punctuation(int32_t uc);
 

data/lib/rinku.rb

@@ -1,5 +1,5 @@
 module Rinku
-  VERSION = "2.0.2"
+  VERSION = "2.0.3"
 
   class << self
     attr_accessor :skip_tags

data/rinku.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name = 'rinku'
-  s.version = '2.0.2'
+  s.version = '2.0.3'
   s.summary = "Mostly autolinking"
   s.description = <<-EOF
     A fast and very smart autolinking library that

data/test/autolink_test.rb

@@ -412,4 +412,22 @@ This is just a test. <a href="http://www.pokemon.com">http://www.pokemon.com</a>
     assert_linked "URL is #{generate_result(url, "mailto:#{url}")}.", "URL is #{url}."
     assert_linked "(URL is #{generate_result(url, "mailto:#{url}")}.)", "(URL is #{url}.)"
   end
+
+  def test_urls_with_parens
+    assert_linked "(<a href=\"http://example.com\">http://example.com</a>)", "(http://example.com)"
+    assert_linked "((<a href=\"http://example.com/()\">http://example.com/()</a>))", "((http://example.com/()))"
+    assert_linked "[<a href=\"http://example.com/()\">http://example.com/()</a>]", "[http://example.com/()]"
+
+    assert_linked "（<a href=\"http://example.com/\">http://example.com/</a>）", "（http://example.com/）"
+    assert_linked "【<a href=\"http://example.com/\">http://example.com/</a>】", "【http://example.com/】"
+    assert_linked "『<a href=\"http://example.com/\">http://example.com/</a>』", "『http://example.com/』"
+    assert_linked "「<a href=\"http://example.com/\">http://example.com/</a>」", "「http://example.com/」"
+    assert_linked "《<a href=\"http://example.com/\">http://example.com/</a>》", "《http://example.com/》"
+    assert_linked "〈<a href=\"http://example.com/\">http://example.com/</a>〉", "〈http://example.com/〉"
+  end
+
+  def test_urls_with_quotes
+    assert_linked "'<a href=\"http://example.com\">http://example.com</a>'", "'http://example.com'"
+    assert_linked "\"<a href=\"http://example.com\">http://example.com</a>\"\"", "\"http://example.com\"\""
+  end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rinku
 version: !ruby/object:Gem::Version
-  version: 2.0.2
+  version: 2.0.3
 platform: ruby
 authors:
 - Vicent Marti
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-10-07 00:00:00.000000000 Z
+date: 2017-08-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake